A Novel Role for the SMG-1 Kinase in Lifespan and Oxidative Stress Resistance in Caenorhabditis elegans

The PTEN tumour suppressor encodes a phosphatase, and its daf-18 orthologue in Caenorhabditis elegans negatively regulates the insulin/IGF-1 DAF-2 receptor pathway that influences lifespan in worms and other species. In order to identify new DAF-18 regulated pathways involved in aging, we initiated a candidate RNAi feeding screen for clones that lengthen lifespan. Here, we report that smg-1 inactivation increases average lifespan in a daf-18 dependent manner. Genetic analysis is consistent with SMG-1 acting at least in part in parallel to the canonical DAF-2 receptor pathway, but converging on the transcription factor DAF-16/FOXO. SMG-1 is a serine-threonine kinase which plays a conserved role in nonsense-mediated mRNA decay (NMD) in worms and mammals. In addition, human SMG-1 has also been implicated in the p53-mediated response to genotoxic stress. The effect of smg-1 inactivation on lifespan appears to be unrelated to its NMD function, but requires the p53 tumour suppressor orthologue cep-1. Furthermore, smg-1 inactivation confers a resistance to oxidative stress in a daf-18-, daf-16- and cep-1-dependent manner. We propose that the role of SMG-1 in lifespan regulation is at least partly dependent on its function in oxidative stress resistance. Taken together, our results unveil a novel role for SMG-1 in lifespan regulation.

In order to identify new DAF-18 partners, we initiated a candidate RNAi feeding screen for clones that lengthen lifespan in a daf-18 dependent manner, focusing on potential protein kinases.
We identified one clone encoding the protein kinase SMG-1. Genetic analyses strongly suggest that SMG-1 acts in parallel to DAF-2, AGE-1 and AKT-1, but requires DAF-16, to modulate lifespan. SMG-1 is conserved across species and is involved in a mechanism responsible for the degradation of premature stop codon containing mRNA, also called NMD for ''nonsense mediated mRNA decay'' in C. elegans and in mammalian cells [14][15][16]. Interestingly the role of SMG-1 in lifespan appears to be unrelated to its function in NMD, but requires the p53 C. elegans ortholog, cep-1, daf-18, and daf-16. Furthermore, our results uncover a role for SMG-1 in oxidative stress response that may be responsible for its effect on lifespan. Overall, our study unveils a novel role for SMG-1 in oxidative stress response and lifespan regulation that may be conserved in mammals.
In addition, from the 269 hand picked clones tested (Table S1) we identified one corresponding to the smg-1 gene. The average lifespan of the rrf-3 strain was increased by 25% when worms were fed either one of the two non-overlapping RNAi clones for smg-1 (C48B6.6 and C48B6.7) and the increase in lifespan was completely suppressed in rrf-3(pk1426); daf-18(e1375) mutants (Table 1; Figure 1A). Therefore, inhibition of smg-1 by RNAi increases lifespan, and this effect requires DAF-18 activity.
We next tested whether the previously isolated smg-1(r861) null allele [18] also shows a lifespan phenotype. The average lifespan of smg-1(r861) mutant animals was not increased compared to wild-type animals (data not shown). Nonetheless, despite the fact that these mutants did not live longer than wild-type, we observed a delayed accumulation of the aging marker lipofuscine [19] during the first week of life of smg-1(r861) mutants, as observed in smg-1 RNAi treated animals (data not shown).
Furthermore, in agreement with previously published data smg-1(r861) null mutants are associated with a fully penetrant protruding vulva phenotype (94%; n = 205), while this phenotype was only observed in 31% (n = 969) of smg-1 RNAi treated animals. The majority of other smg-1 mutants we tested behave like the smg-1(r861) null allele (data not shown) besides smg-1(tm869) mutants (recently isolated by the Japanese C.elegans knockout consortium). An exception is the smg-1(tm869) allele, which results in a protruding vulva phenotype with similar penetrance (47%, n = 351) to smg-1 RNAi fed animals. Indeed, lifespan tests revealed that the smg-1(tm869) mutation increases average lifespan by more than 20% ( Figure 1B and Table 1). Overall, our results suggest that RNAi mimics a hypomorphic mutation, while complete loss of SMG-1 function is deleterious and masks a longevity phenotype.

SMG-1 may act in parallel of the insulin/IGF-1 DAF-2 receptor
Since DAF-18 functions in the insulin/IGF-1 DAF-2 receptor pathway to regulate lifespan, we tested whether SMG-1 also acts in this signaling cascade.
We favored RNAi approaches to assess epistatic relationships between smg-1 and the different components of the insulin pathway in order to analyse data in an isogenic background. RNAi of daf-2 increased lifespan by 84% compared to control RNAi (Table 1; Figure 2B). RNAi of both daf-2 and smg-1 further extended the average lifespan to 120% (Table 1; Figure 2B). Similarly, the average lifespan of age-1 RNAi and akt-1 RNAi treated worms was further increased from 40 to 84% and from 50 to 75%, respectively, when fed with smg-1 RNAi (Table 1; Figure 2 C,D). Conversely, the lifespan of animals treated with RNAi for both daf-2 and age-1, which act in the same pathway, was not significantly different from the lifespan of daf-2 RNAi worms alone (Table 1; Figure 2A). Therefore, smg-1 inactivation increases lifespan independently of daf-2, age-1 or akt-1. However, the extension of lifespan by smg-1 RNAi was completely suppressed when daf-16 was inactivated by RNAi (Table 1; Figure 2E). Overall, these data suggest that SMG-1 may act in a pathway parallel to DAF-2, AGE-1 and AKT-1, but requiring DAF-16 activity.
Nonetheless, because gene inactivation by RNAi mimics a hypomorphic rather than a null mutation, we cannot formally exclude that the insulin receptor pathway partially contributes to the smg-1 effect on lifespan.
DAF-16 function can be modulated through its nuclear localization [6,7,20,21]. To investigate whether SMG-1 controls DAF-16 sub-cellular localization, we made use of a strain carrying a daf-16::gfp reporter construct to visualize nuclear translocation in vivo [6]. DAF-16::GFP was localized in both the cytoplasm and the nucleus in all tissues of worms after smg-1 inactivation by RNAi or by mutation (Figure 3), as observed in control worms. Conversely, under the same experimental conditions, daf-2 RNAi induced DAF-16::GFP nuclear accumulation ( Figure 3). These results strongly suggest that SMG-1 does not regulate DAF-16 activity through its sequestration into the cytoplasm, and further support the idea that SMG-1 and DAF-2 may act in different pathways to regulate lifespan.
Nonetheless, these observations do not exclude that SMG-1 may behave as a weak enhancer of the insulin pathway, since Henderson and Johnson [6] showed that some age-1 mutants also fail to induce nuclear re-localization of this reporter. We thus addressed the relationship between the DAF-2 pathway and SMG-1 by a third approach.
The DAF-2 pathway is also critical for controlling dauer formation. To further assess a potential functional link between SMG-1 and DAF-2 we tested the involvement of SMG-1 in dauer formation. One would predict that if the role of SMG-1 in lifespan control relies on its interaction with the DAF-2 pathway, then SMG-1 inhibition should increase dauer formation. However, we found that smg-1 RNAi slightly increased the ability of worms to recover from larval arrest rather than enhancing dauer formation. When daf-2(e1370) or daf-2(e1370); rrf-3 (pk1426) double mutants were maintained at the semi-nonpermissive temperature 22uC [22], 78+/27% of smg-1 RNAi treated worms had reached the adult stage after 4 days, compared to 68+/24.8% of worms grown on HT115 control bacteria (P = 0.02). These data strongly suggest that SMG-1 is not a broad positive regulator of the insulin pathway.
Overall our data support a model in which SMG-1 functions at least in part independently of DAF-2 to regulate lifespan.
The role of SMG-1 in lifespan control does not depend on its function in NMD smg-1 encodes a conserved serine threonine kinase involved in nonsense mediated mRNA decay (NMD), a mechanism responsible for the degradation of mRNAs containing a premature stop codon [14][15][16]. In addition to SMG-1, six evolutionary conserved SMG proteins are also required for NMD in C. elegans and in mammalian cells. Genetic studies have determined that smg genes are regulators of the phosphorylation state of SMG-2. SMG-1, SMG-3 and SMG-4 are required for the phosphorylation of SMG-2, whereas SMG-5, SMG-6 and SMG-7 are involved in its dephosporylation [23].
If the role of SMG-1 in lifespan relies on its function in NMD, inactivation of other smg genes should also have an effect on lifespan. The average lifespan of worms fed with smg-2, smg-4, smg-5 or smg-7 RNAi clones was not significantly increased ( Table 1), suggesting that NMD inactivation may not be responsible for lifespan extension. To further explore this hypothesis, we assess NMD activity in smg-1 RNAi treated worms. Longman et al. [24] developed an assay using transgenic strains expressing a GFP reporter constructs either with a natural stop codon or harboring a premature termination codon (PTC). Introduction of a PTC into the reporter induces a robust NMD response, as determined by the lack of GFP expression in transgenic worms. Conversely, smg-2 RNAi, which abrogates NMD, restores GFP expression [24]. Under our experimental conditions, GFP expression was induced in 100% of PTC transgenic worms fed with the smg-1 RNAi clone (Figure 4), demonstrating the effectiveness of smg-1 RNAi feeding in NMD inhibition. daf-18 and daf-16 are required for the smg-1 dependent lifespan increase. We reasoned that if NMD inhibition is responsible for the role of smg-1 in lifespan control, inactivation of daf-18 or daf-16 should antagonize this function and thus impede GFP expression in smg-1 RNAi treated PTC transgenic worms. Inactivation of either daf-18 or daf-16 by RNAi, which is sufficient to suppress the smg-1 lifespan phenotype, did not reduce Overall our results show that there is no correlation between NMD inactivation and lifespan phenotypes, thus indicating that SMG-1 functions independently of NMD to regulate lifespan.

p53/cep-1 is involved in smg-1 dependent lifespan modulation
It was recently reported that human SMG-1 is functionally linked to the tumor suppressor checkpoint protein p53. hSMG-1 phosphorylates and stabilizes p53 in response to genotoxic stress induced by UV and c irradiation [25]. In worms, the p53 homologue cep-1 is required for DNA damage-induced apoptosis [26,27]. Interestingly, Arum et al. recently showed that cep-1 mutations also increase longevity without altering DAF-16::GFP nuclear localization [28] We therefore tested whether cep-1 is involved in the regulation of lifespan by smg-1. cep-1 RNAi partially suppressed the extension of lifespan due to smg-1 inhibition (Table 1; Figure 5). The genetic interaction between cep-1 and smg-1 is specific, as cep-1 RNAi alone did not reduce lifespan (Table 1; Figure 5). These results indicate that when smg-1 is inactivated, cep-1 is required to extend lifespan.

smg-1 inactivation confers resistance to oxidative stress
Resistance to oxidative stress is a hallmark of many longevity mutants in C. elegans [29]. Consistently, we observed that worms were more resistant to oxidative stress induced by paraquat when smg-1 was inactivated by RNAi or by mutation, as 47% of worms were still alive after 7 days of paraquat treatment compared to 7% for control RNAi (Figure 6 A, B). Conversely, daf-18 and daf-16 RNAi inhibited the resistance of worms to oxidative stress compared to control RNAi and dramatically reduced the stress resistance induced by smg-1 inactivation (Figure 6 A, B). These results show that SMG-1 requires DAF-16 and DAF-18 to confer oxidative stress resistance as well as to function in lifespan control. A correlation between lifespan and oxidative stress resistance phenotypes was also observed for the genetic interaction between daf-2 and smg-1, as the resistance to oxidative stress of daf-2 RNAi treated animals was further increased by smg-1 inactivation (Figure 6 A, B; p,10 23 ).
To further explore the relationship between lifespan and resistance to oxidative stress phenotypes, we investigated the role of cep-1/p53 in oxidative stress resistance. Under experimental conditions where cep-1 RNAi increased lifespan ( Figure 5), resistance to oxidative stress of RNAi treated animals increased  significantly compared to control animals (Figure 6 C, D; p,10 23 ). Conversely, the stress resistance of smg-1 RNAi animals was reduced by cep-1 RNAi (Figure 6 C, D; p,10 23 ), as expected from the partial suppression observed for the lifespan phenotype ( Figure 6).
Altogether, our results show that CEP-1 has opposing effects on oxidative stress resistance depending on the presence or absence of smg-1. This apparent paradox may be explained by the role of p53 in mammalian cells, where it displays either pro-oxidant or antioxidant functions depending on the level of oxidative stress (high or low, respectively, [30]. In wild-type animals treated with paraquat, where the level of oxidative stress is high, p53 plays a pro oxidant function, thus explaining the beneficial effect of reducing its expression. Upon smg-1 inactivation, the levels of oxidative stress may be lower; p53 could play an antioxidant function in this context [30], explaining the deleterious effect of its removal. Overall, our results show that positive and negative regulators of SMG-1 activity in lifespan regulation act in a similar manner in the oxidative stress response. These data support the hypothesis that the resistance to oxidative stress of smg-1 animals may be responsible for  their increased lifespan phenotype. However, other mechanisms may also be involved since daf-16 (or daf-18) inactivation is sufficient to suppress smg-1 lifespan regulation without completely inhibiting the smg-1 stress resistance phenotype.

A role for SMG-1 in sensory neuron signaling for lifespan regulation
Several observations prompted us to investigate the role of smg-1 in the regulation of lifespan via sensory neurons. Firstly, among the different mechanisms that modulate C. elegans lifespan, mutations in sensory neurons lengthen lifespan in a daf-16 dependent manner [31]. Secondly, a number of genes that act in the nervous system have been shown to be refractory to RNAi in a wild type context, but efficiently inactivated in an rrf-3 RNAi hyper-sensitive background [17]. Similarly, smg-1 RNAi increased lifespan in RNAi hyper-sensitive backgrounds such as rrf-3 and ppw-1 [32], but not in wild-type animals (Figure 1 and data not shown).
daf-19, which encodes an evolutionary conserved RFX-type transcription factor, is a master gene for the development of a ciliary module in C. elegans [33]. In order to assess the role of SMG-1 in sensory neuron formation, animals were stained with DiO, a fluorescent probe which enters through functional cilia (see material and methods).We counted on average 10.2+/20.2 (n = 50); 9.4+/20.3 (n = 42); 9.6+/20.2 (n = 29) and 0 (n = 21) ciliated neurons, respectively, in rrf-3 animals maintained on control bacteria, rrf-3 animals on daf-19 RNAi, in daf-12(sa204) mutants, and in daf-19(m86); daf-12(sa204) mutants. Therefore, while daf-19(m86) mutants do not stain at all [33], inhibition of daf-19 expression by RNAi is not strong enough to interfere with sensory neurons development, as also observed by others (Peter Swoboda, personal communication). Furthermore, smg-1 RNAi treated animals stained a similar number of sensory neurons as animals fed on control bacteria (10.5+/20.2, n = 40). Overall our results show that smg-1 inhibition by RNAi does not compromise sensory neurons formation. These results do not however exclude a possible function for SMG-1 in sensory perception. The smg-1 predicted promoter contains the canonical DAF-19 target sequence [34], suggesting that smg-1 expression may be regulated by DAF-19. As daf-19 expression is restricted to sensory neurons, smg-1 may well play a role in these cells to regulate lifespan. Nonetheless, the SMG-1 mode of action differs from the previously described pathways, as sensory mutants, but not smg-1 inactivation, affect DAF-16 nuclear translocation [31].
As observed upon smg-1 inactivation, resistance to oxidative stress was increased after daf-19 RNAi, and was suppressed when daf-19 was inactivated in combination with daf-16 or daf-18 ( Figure 7C, D). Furthermore, in contrast to daf-2 and smg-1 double RNAi, inactivation of daf-19 did not confer higher resistance to smg-1 RNAi animals. These results strongly suggest that SMG-1 may function with DAF-19 to regulate both lifespan and oxidative stress.
In conclusion, we identified smg-1 as a novel gene involved in lifespan regulation. Furthermore, our results suggest that SMG-1 may act independently of DAF-2 and requires DAF-18/PTEN, DAF-16/FOXO and CEP-1/p53 to regulate lifespan.
In mammalian cells, p53 and FOXO3A can act as cofactors to regulate transcription [35]. Furthermore, PTEN has also been shown to interact with and to stabilize the p53 protein [36]. Thus, it is tempting to speculate that SMG-1 may affect DAF-16 transcriptional activity via the regulation of PTEN and p53. Interestingly, the physical interaction between p53 and PTEN has recently been shown to be regulated by oxidative stress [37], and their functional crosstalk does not require the lipid phosphatase activity of PTEN [37]. This is in agreement with our results strongly suggesting that SMG-1 acts independently of the PI 3 kinase AGE-1. Finally, pull down experiments revealed that PTEN and hSMG-1 physically interact in human cells [38]. It is therefore possible that mammalian ortholog of smg-1 control lifespan and the response to oxidative stress in mammals. Recent data suggests a function for SMG-1 both in the oxidative stress response [39] and a role in apoptosis unrelated to the suppression of nonsense-mediated mRNA decay [40]. Understanding the molecular interactions and mechanisms of the pathway involving SMG-1 in aging will be the challenge for future studies.
Double RNAi experiments were carried out by mixing the bacterial cultures directly before seeding the NGM plates. Controls were RNAi clone 50% diluted with vector control RNAi bacteria.

Lifespan assays
Animals were grown on regular NGM plates at 20uC until reaching the L4 stage and then transferred to RNAi plates (F0). F0 adults were removed after 24h and F1 L4 were transferred to 10 mM 5-fluorodeoxyuracile (5-FU, Sigma-Aldrich, Steinheim, Germany) containing plates to prevent growth of progeny. Lifespan assays were performed at 20uC. The day of the shift is counted as day 0 in the adult lifespan assay. Control and experimental animals were transferred in parallel to fresh RNAi plates once a week. Lifespan was assessed every 2-3 days and animals were scored as dead when they ceased moving and responding to prodding. Animals that crawled off the plate, had a ''protruding vulva'' or an ''exploded vulva'' phenotype were censored. smg-1 RNAi was also performed in absence of 5-FU and gave similar results (data not shown).
Survival analyses were performed using the Kaplan Meier method and the significance of differences between survival curves calculated using the log rank test. The statistical software used was SPSS, Version 11.5 (SPSS, Chicago, IL, USA) and all Pvalues,0.05 were considered significant.

Assessment of NMD activity in living worms
Animals carrying the PTC transgenic reporter [24] were fed at 20uC with the control clone only or with the smg-1 RNAi clone mixed either with the control, or daf-18, or daf-16 RNAi clones. F1 animals were scored for GFP expression at the L4 stage. For GFP intensity quantification, animals were photographed under a GFP filter and the average brightness was determined for each photograph by Lucia Nikon software. All images were handled identically. At least 30 animals per RNAi conditions were averaged.

Stress resistance assays
Synchronously cultured animals were kept on NGM plates at 20uC until the young adult stage. For each strain, 5 6 20 young adults were transferred on Paraquat (methylviologene, Sigma-Aldrich, Steinheim, Germany) containing plates (90 ml of 150 mM Paraquat added on top of NGM plates already seeded with HT115 bacteria). Surviving animals were scored every day during 8-9 days. At least three independent RNAi experiments have been conducted for each clone tested. P-values were calculated using the t-Student test to determine differences in oxidative stress resistance.

Larval arrest assays
Five young adults daf-2(e1370) ; rrf-3(pk1426) double mutants were fed at 22uC with either HT115 or smg-1 RNAi bacteria, then removed 24 hours later. F1 progenies were followed every day and the numbers of worms that have reached the adult stage were counted at day 4. Numbers are given for 3 independent experiments.
Observation of DAF-16::GFP sub-cellular localization and DiO staining The sub-cellular localization of the DAF-16::GFP protein was analyzed in the smg-1(r861) mutant background and after smg-1 RNAi inactivation in a rrf-3(pk1426) mutant background by fluorescence microscopy under a GFP filter. About 10 worms were mounted on agar pads (2% agarose with 5 mM tetramisole) to avoid DAF-16::GFP translocation due to stress [8]. At least 20 animals were examined for each developmental stage (embryo, L1, L2, L3, L4 and adult). Sensory neurons were stained by incubating L4 worms in M9 containing DiO (Molecular Probes) at 10 mg/ml final concentration for two hours. Worms were then transfered to plates for one hour and observed by fluorescence microscopy under a GFP filter.